Jinlong Xie

Under the mounting pressure of global green transitions (e.g., the EU CBAM), resource-constrained manufacturing SMEs are rapidly adopting Generative AI to handle complex carbon accounting and compliance audits. While this appears to drive a surge in compliance efficiency and knowledge externalization, my research uncovers the hidden crisis behind this "digital prosperity."

By conceptualizing Generative AI as a "cognitive prosthesis," this study reveals how the delegation of core epistemic judgments to inscrutable AI algorithms disrupts the transactive memory systems (TMS) within organizations. I identify a tragic decoupling paradox: the use of AI grants SMEs immediate short-term compliance legitimacy but severely paralyzes their long-term dynamic combinative capabilities. Furthermore, my research introduces "adversarial epistemic governance" (human-in-the-loop review mechanisms) as a critical boundary condition to prevent organizational capability atrophy, offering a novel perspective on human-AI interaction in high-stakes routines.